Research on vision-based autonomous precise landing of unmanned aerial vehicles and adaptive inspection software for UAVs

As unmanned aerial vehicle (UAV) inspection technology becomes increasingly prevalent in routine transmission line inspection tasks, this paper addresses the limitations of traditional UAV navigation systems, including low precision, significant adaptive inspection errors, and difficulties in identifying and tracking inspection targets. To address these issues, a vision-based autonomous precise landing and adaptive inspection system for UAVs is proposed. Firstly, a precise landing marker is designed utilizing vision navigation technology. An optimization model for UAV pose recognition is constructed based on the mapping relationship of coordinate systems, where feature point coordinates serve as the foundation for algorithm-based relative pose estimation between landing markers, thereby enhancing the precision of autonomous precise landing. Additionally, to improve the accuracy of adaptive inspection positioning and tracking for UAVs, an optimal threshold selection algorithm based on changing scenarios is proposed. This algorithm enables precise predictive and continuous parameter selection, further enhancing the accuracy of target localization and autonomous tracking during inspections. A robust adaptive federated Kalman filter tracking algorithm is employed for adaptive positioning recognition and detection comparison. The proposed scheme exhibits significantly higher precision than the other two alternatives, with an 83% improvement in accuracy. This paper's vision-based adaptive inspection technology for UAVs, integrated with robust adaptive filtering and tracking algorithms, effectively resolves the issues of low precision and tolerance in traditional navigation systems. It enhances the intelligence and finesse of UAV adaptive inspection technology